1 00:00:00,000 --> 00:00:07,280 My name is Keegan. 2 00:00:07,280 --> 00:00:14,500 Got into the cannabis space in 2018 at a laboratory and have now been working with cannabis data 3 00:00:14,500 --> 00:00:16,820 going on for plus years now. 4 00:00:16,820 --> 00:00:22,200 So love to hear about what you'd like in the industry, what you're doing and what you hope 5 00:00:22,200 --> 00:00:23,520 to get out of the group. 6 00:00:23,520 --> 00:00:26,600 So Daniel, would you mind kicking it off? 7 00:00:26,600 --> 00:00:27,600 Sure. 8 00:00:27,600 --> 00:00:35,360 I just completed a data analytics program over the past few months and was looking at 9 00:00:35,360 --> 00:00:38,840 different types of data events and this popped up at the top. 10 00:00:38,840 --> 00:00:40,480 So I figured I'd come check it out. 11 00:00:40,480 --> 00:00:41,480 Certainly. 12 00:00:41,480 --> 00:00:42,480 My name is Dean Pangelinan. 13 00:00:42,480 --> 00:00:46,000 I'm a senior business intelligence strategist for a company called Merge World. 14 00:00:46,000 --> 00:00:51,280 We're in the digital marketing space and some of our clients are actually in the cannabis 15 00:00:51,280 --> 00:00:52,280 industry. 16 00:00:52,280 --> 00:00:59,960 So I'm wanting to learn more about the scientific analysis of cannabis data because I am a statistician 17 00:00:59,960 --> 00:01:05,100 by training and work primarily in the visualization and representation of the data. 18 00:01:05,100 --> 00:01:11,040 So I want to be able to find out how I can best help our clients and candidates. 19 00:01:11,040 --> 00:01:12,960 You're definitely in the right place. 20 00:01:12,960 --> 00:01:16,560 Data visualization is a key part of what we do. 21 00:01:16,560 --> 00:01:19,880 So would love to have your expertise. 22 00:01:19,880 --> 00:01:22,400 I'm thrilled to pick your brain. 23 00:01:22,400 --> 00:01:26,120 TJ, would you be interested in introducing yourself? 24 00:01:26,120 --> 00:01:27,120 Sure. 25 00:01:27,120 --> 00:01:28,720 Hi, I'm TJ Goff. 26 00:01:28,720 --> 00:01:34,560 I'm a back end engineer working at Lantern, which is a cannabis delivery startup based 27 00:01:34,560 --> 00:01:36,960 out of Boston. 28 00:01:36,960 --> 00:01:37,960 Too cool. 29 00:01:37,960 --> 00:01:40,360 Cannabis delivery is a big, big thing. 30 00:01:40,360 --> 00:01:43,720 So you're well positioned for a profitable future. 31 00:01:43,720 --> 00:01:44,800 So awesome. 32 00:01:44,800 --> 00:01:45,800 That's the idea. 33 00:01:45,800 --> 00:01:50,560 We'll have to pick your brain on transportation because we've got a lot of transportation 34 00:01:50,560 --> 00:01:51,560 data. 35 00:01:51,560 --> 00:01:55,760 Ramni, would you be interested in introducing yourself? 36 00:01:55,760 --> 00:01:56,760 Yes. 37 00:01:56,760 --> 00:01:59,840 Hi, thank you for organizing this meetup. 38 00:01:59,840 --> 00:02:05,280 So I'm taking this program in data analytics and I'm learning all the machine learning 39 00:02:05,280 --> 00:02:10,120 and statistical modeling and data visualization and all the different techniques. 40 00:02:10,120 --> 00:02:14,400 So when I saw this, I found the data is interesting, like cannabis data. 41 00:02:14,400 --> 00:02:15,400 Wow. 42 00:02:15,400 --> 00:02:16,880 I've never seen that data. 43 00:02:16,880 --> 00:02:20,960 So hence I'm joining and let's see what I learned from here. 44 00:02:20,960 --> 00:02:21,960 Thank you. 45 00:02:21,960 --> 00:02:22,960 Awesome. 46 00:02:22,960 --> 00:02:25,200 You're also in the right place. 47 00:02:25,200 --> 00:02:26,200 Thanks for joining. 48 00:02:26,200 --> 00:02:30,320 Trav, would you be interested in introducing yourself? 49 00:02:30,320 --> 00:02:31,640 Yes, indeed. 50 00:02:31,640 --> 00:02:32,640 I'm Trav. 51 00:02:32,640 --> 00:02:36,880 I'm just interested in the industry and I want to learn more about data analytics. 52 00:02:36,880 --> 00:02:37,880 Too cool. 53 00:02:37,880 --> 00:02:39,960 It's going to be a big field. 54 00:02:39,960 --> 00:02:44,240 So once again, there's a high demand for people with data analytics skills. 55 00:02:44,240 --> 00:02:47,640 So you're well positioned too. 56 00:02:47,640 --> 00:02:53,320 Well, without further ado, I'll go ahead and share with you some of the work I've been 57 00:02:53,320 --> 00:03:01,340 doing because really it's just to guide a discussion. 58 00:03:01,340 --> 00:03:07,600 So don't let me steal the show, but hopefully it'll spark a good conversation here. 59 00:03:07,600 --> 00:03:09,840 Well, welcome to the meetup. 60 00:03:09,840 --> 00:03:12,800 Last week we looked at laboratories. 61 00:03:12,800 --> 00:03:18,040 And then this week I figured let's get a bit more to the production side. 62 00:03:18,040 --> 00:03:24,400 We'll get to cultivation next week, but we can start with processing. 63 00:03:24,400 --> 00:03:29,600 So I'm sure all of you are acquainted with processed cannabis. 64 00:03:29,600 --> 00:03:38,300 Up in the top left corner, we've got product known as Shatter testing at 74% THC. 65 00:03:38,300 --> 00:03:45,000 And so this is a new and an old process. 66 00:03:45,000 --> 00:03:50,080 So I was going to dive in a bit to the history before we get to the data because it's so 67 00:03:50,080 --> 00:03:56,240 fascinating to know where all these processes came from. 68 00:03:56,240 --> 00:04:05,200 So I said, you know, humans are basically just remarkably good at using plants as factories 69 00:04:05,200 --> 00:04:08,320 to produce complex chemicals. 70 00:04:08,320 --> 00:04:14,240 And people have been doing this with cannabis hemp for a long time now. 71 00:04:14,240 --> 00:04:22,240 So people think that cannabis was being processed in India as early as 1000 BC. 72 00:04:22,240 --> 00:04:30,200 So we're going on potentially 3,000 years of people processing and using cannabis. 73 00:04:30,200 --> 00:04:35,880 And they may not have had the high THC that we had today, so that may be why it was important 74 00:04:35,880 --> 00:04:39,340 to produce concentrates. 75 00:04:39,340 --> 00:04:43,080 So just fascinating history here. 76 00:04:43,080 --> 00:04:46,960 And we'll dive into some of the more modern history. 77 00:04:46,960 --> 00:04:55,600 So what we all know now as cannabis brownies, and sorry you got cut off here, we've got 78 00:04:55,600 --> 00:05:01,320 Alice B. Topeless, and I'll send out a reformatted presentation. 79 00:05:01,320 --> 00:05:08,200 But she was the first one to really write a cookbook with cannabis in chocolate. 80 00:05:08,200 --> 00:05:12,480 So you know, you can consider that processing. 81 00:05:12,480 --> 00:05:22,120 And the recipe was by Brian Dyson, who was a restaurateur. 82 00:05:22,120 --> 00:05:31,240 So interesting history there. 83 00:05:31,240 --> 00:05:34,600 So they were cooking with cannabis, concentrating it. 84 00:05:34,600 --> 00:05:41,520 And then beginning in 1940 and really accelerating into the 1960s is when people really started 85 00:05:41,520 --> 00:05:44,440 isolating the various cannabinoids. 86 00:05:44,440 --> 00:05:51,800 So you know, when you look at the grand scheme of things, we've only been, you know, concentrating 87 00:05:51,800 --> 00:06:01,680 these products and really, you know, analyzing them on a compound by compound basis for 60 88 00:06:01,680 --> 00:06:08,880 or so years, maybe 80, but about 60, you know, rigorously, scientifically. 89 00:06:08,880 --> 00:06:14,000 So that's pretty exciting that, you know, cannabis has been around, you know, 3000 years 90 00:06:14,000 --> 00:06:15,420 or so. 91 00:06:15,420 --> 00:06:21,240 And now we get to really get scientific at a real molecular level. 92 00:06:21,240 --> 00:06:30,120 So it's a real exciting, real exciting work, you know, still to be done. 93 00:06:30,120 --> 00:06:35,460 I'll kind of skim over this, but you know, these days just wanted to talk about how cannabis 94 00:06:35,460 --> 00:06:37,500 extraction is done. 95 00:06:37,500 --> 00:06:40,820 A lot of times it's done with organic solvents. 96 00:06:40,820 --> 00:06:47,320 You can do solventless, which is anecdotally becoming more popular. 97 00:06:47,320 --> 00:06:53,200 So some of the solvents that you may use are hydrocarbons. 98 00:06:53,200 --> 00:06:59,560 You also can use butane and carbon dioxide. 99 00:06:59,560 --> 00:07:06,720 But you know, you've got to be careful, right, because not only are these chemicals flammable, 100 00:07:06,720 --> 00:07:10,600 but you don't necessarily want to be ingesting them either. 101 00:07:10,600 --> 00:07:17,760 So there's, you know, a purging process where, you know, you try to get rid of all these 102 00:07:17,760 --> 00:07:18,760 solvents. 103 00:07:18,760 --> 00:07:24,540 And then, you know, we see the data at the laboratory where not only are you testing 104 00:07:24,540 --> 00:07:30,840 for CBD, THC, you also want to make sure you don't have these residual solvents left in 105 00:07:30,840 --> 00:07:34,240 your product. 106 00:07:34,240 --> 00:07:40,980 And in particular, what we'll be talking about today is processing variability. 107 00:07:40,980 --> 00:07:51,240 So we'll be looking at the various compounds that producers may be trying to isolate. 108 00:07:51,240 --> 00:07:56,260 And we'll look at essentially how good they are at doing that. 109 00:07:56,260 --> 00:08:05,280 So that was our original question is, you know, how to run a profitable processor. 110 00:08:05,280 --> 00:08:11,380 And one can argue that focusing on variability could be key. 111 00:08:11,380 --> 00:08:17,100 So for financial reasons, you're always trying to minimize risk. 112 00:08:17,100 --> 00:08:26,480 And in finance, you often just use risk as a synonym for, you know, variability, you 113 00:08:26,480 --> 00:08:29,440 know, your standard deviations in returns. 114 00:08:29,440 --> 00:08:36,160 So from a financial point of view, which drives a lot of the action here in the cannabis industry. 115 00:08:36,160 --> 00:08:39,780 So also for economic reasons, people are risk averse. 116 00:08:39,780 --> 00:08:48,280 So people are just predisposed to processes that aren't unpredictable. 117 00:08:48,280 --> 00:08:54,280 From an engineering point of view, you want to minimize your variance because it's going 118 00:08:54,280 --> 00:08:56,320 to be easier to design your system. 119 00:08:56,320 --> 00:09:05,320 It's going to be much harder to design your process if you've got variation in your extraction 120 00:09:05,320 --> 00:09:07,280 techniques. 121 00:09:07,280 --> 00:09:12,560 And then, of course, from a business marketing and public health perspective, you know, you 122 00:09:12,560 --> 00:09:16,080 want to maximize your consistency of your product. 123 00:09:16,080 --> 00:09:22,960 So marketers are anecdotally talking about how they want a consistent product for their 124 00:09:22,960 --> 00:09:24,640 customers. 125 00:09:24,640 --> 00:09:31,840 And then from a public health point of view, if people are using these concentrates for 126 00:09:31,840 --> 00:09:41,040 medicinal purposes or what have you, you know, you want to make sure that people are, you 127 00:09:41,040 --> 00:09:44,400 know, consuming what they think they're going to be consuming. 128 00:09:44,400 --> 00:09:49,600 So that's where, you know, consistency, you know, matters from, you know, a public health 129 00:09:49,600 --> 00:09:52,480 point of view. 130 00:09:52,480 --> 00:10:00,680 And for the newcomers, I'll let you introduce yourselves here in just a minute. 131 00:10:00,680 --> 00:10:08,280 I'll talk about the data in the code here, and then we can reintroduce ourselves for 132 00:10:08,280 --> 00:10:15,960 the newcomers and then just have open discussions. 133 00:10:15,960 --> 00:10:21,240 And please interrupt me if I'm, you know, moving a little fast here. 134 00:10:21,240 --> 00:10:23,720 Awesome. 135 00:10:23,720 --> 00:10:28,560 Well, on to the data. 136 00:10:28,560 --> 00:10:35,800 So always working with open source data here. 137 00:10:35,800 --> 00:10:40,160 And I'll share these links with you afterwards. 138 00:10:40,160 --> 00:10:47,720 But the Washington state has a very liberal Freedom of Information Act. 139 00:10:47,720 --> 00:10:57,080 And so entities like the Cannabis Observer will do periodic Freedom of Information Act 140 00:10:57,080 --> 00:11:02,760 requests to get access to this publicly available data. 141 00:11:02,760 --> 00:11:09,880 And so here is a release from November of 2021. 142 00:11:09,880 --> 00:11:12,160 So still quite recent data. 143 00:11:12,160 --> 00:11:19,520 And this data spans from 2018 till November of 2021. 144 00:11:19,520 --> 00:11:21,640 Caveat. 145 00:11:21,640 --> 00:11:25,200 This data can be notoriously hard to work with. 146 00:11:25,200 --> 00:11:33,360 So for example, we're looking at the lab results and, you know, unzipped, you know, the lab 147 00:11:33,360 --> 00:11:36,820 results can be quite sizable. 148 00:11:36,820 --> 00:11:53,120 So I've downloaded the these data sets here. 149 00:11:53,120 --> 00:11:59,720 We have clocking in at 31 gigabytes. 150 00:11:59,720 --> 00:12:04,920 We have the inventories data weighing in at 14 gigabytes. 151 00:12:04,920 --> 00:12:08,700 We've got the inventory types data. 152 00:12:08,700 --> 00:12:17,140 And then we have the various lab results data, which is about two and a half gigabytes total. 153 00:12:17,140 --> 00:12:26,100 So we have almost 50 gigabytes of data here just pertaining to lab results. 154 00:12:26,100 --> 00:12:33,920 So that can be a daunting task to get your mind around. 155 00:12:33,920 --> 00:12:44,000 But we've written a useful script here, which I'll actually go ahead and share with the 156 00:12:44,000 --> 00:12:50,440 group here. 157 00:12:50,440 --> 00:12:52,360 Probably should have committed this earlier. 158 00:12:52,360 --> 00:13:01,080 But here's the latest code for today's meetup, which you can find on the cannabis data science 159 00:13:01,080 --> 00:13:03,440 GitHub repository. 160 00:13:03,440 --> 00:13:13,820 So for those of you who are new, you can find all the source code here in the cannabis 161 00:13:13,820 --> 00:13:15,980 data science repository. 162 00:13:15,980 --> 00:13:20,760 So that way you can follow along from the code for today. 163 00:13:20,760 --> 00:13:25,640 And so today the code was quite long. 164 00:13:25,640 --> 00:13:29,940 So I'm not going to necessarily go through it line by line. 165 00:13:29,940 --> 00:13:35,760 But if you need Python snippets for how to parse data, then this is a useful script. 166 00:13:35,760 --> 00:13:41,880 But basically what I do is I read in all of these lab results. 167 00:13:41,880 --> 00:13:56,360 And just to give you an idea of some of the data points we're working with here. 168 00:13:56,360 --> 00:14:06,240 For each lab result, we have a handful of cannabinoids, solvents, and then of course, 169 00:14:06,240 --> 00:14:15,840 some of your other screenings, such as for foreign matter, microbes, moisture, and microtoxin, 170 00:14:15,840 --> 00:14:19,040 as well as some other IDs. 171 00:14:19,040 --> 00:14:24,440 We can then match this with the licensees data. 172 00:14:24,440 --> 00:14:29,320 So we can get fields such as the license number. 173 00:14:29,320 --> 00:14:34,360 We can even get things like the city and address. 174 00:14:34,360 --> 00:14:39,840 So zip code of where the licensee may be. 175 00:14:39,840 --> 00:14:47,160 So that way you can even do geographic analysis. 176 00:14:47,160 --> 00:14:51,320 Also have a rich inventories fields. 177 00:14:51,320 --> 00:14:55,360 So here we can get just a couple things of interest. 178 00:14:55,360 --> 00:14:59,560 So in a future work, I want to look at quantities. 179 00:14:59,560 --> 00:15:02,600 So the quantities being produced. 180 00:15:02,600 --> 00:15:06,400 Of course, we're looking at inventory types. 181 00:15:06,400 --> 00:15:11,440 And this is where we get a lot of our IDs. 182 00:15:11,440 --> 00:15:18,920 Then we have to match this with the inventory type to find out what type of product it is. 183 00:15:18,920 --> 00:15:28,720 And then finally, we can get the name of the strain that people gave their product. 184 00:15:28,720 --> 00:15:33,040 So this script essentially does that. 185 00:15:33,040 --> 00:15:44,600 So reads in your lab data, combines it with inventory data, combines it with inventory 186 00:15:44,600 --> 00:15:51,480 type data, combines the lab results with strain data. 187 00:15:51,480 --> 00:15:57,240 Then you combine the lab results with licensee data. 188 00:15:57,240 --> 00:16:03,360 And then you get all of the lab result fields that we're interested in here. 189 00:16:03,360 --> 00:16:06,960 So here are all of our fields we've collected to this point. 190 00:16:06,960 --> 00:16:09,800 A lot of identification fields. 191 00:16:09,800 --> 00:16:17,080 And then we add on all the cannabinoids and solvents and everything else that we would 192 00:16:17,080 --> 00:16:19,520 like. 193 00:16:19,520 --> 00:16:22,420 And I'll let you play around with this script. 194 00:16:22,420 --> 00:16:26,280 But the exciting thing is the data. 195 00:16:26,280 --> 00:16:33,240 So here we started with almost 50 gigabytes of data. 196 00:16:33,240 --> 00:16:48,120 On my first pass, I just shaved it down to the bare minimum with 30 megabytes of IDs 197 00:16:48,120 --> 00:16:52,680 and two gigabytes of names. 198 00:16:52,680 --> 00:16:56,180 And then you can start to clean that out. 199 00:16:56,180 --> 00:17:02,880 So if you add the IDs with the licensee data, actually first you add it with the strain 200 00:17:02,880 --> 00:17:04,160 name data. 201 00:17:04,160 --> 00:17:07,280 And then you're around 40 megabytes. 202 00:17:07,280 --> 00:17:12,440 You add that with the licensees data and you're around 90 megabytes. 203 00:17:12,440 --> 00:17:16,200 And then you add back the lab results data. 204 00:17:16,200 --> 00:17:26,160 And you're sitting less than 200 megabytes, which is something to be happy with when you're 205 00:17:26,160 --> 00:17:31,360 starting out with 50 gigabytes of data. 206 00:17:31,360 --> 00:17:40,020 So that just shows you how much cleaning can reduce the fingerprint of your data. 207 00:17:40,020 --> 00:17:48,100 So here we just stripped out all the extraneous fields and we removed all the null values. 208 00:17:48,100 --> 00:17:58,600 And what we're left with is a large data set, but less of a size than some apps that are 209 00:17:58,600 --> 00:18:01,260 out there. 210 00:18:01,260 --> 00:18:07,320 So it's still large, but it's something that we can work with here. 211 00:18:07,320 --> 00:18:17,160 So just to give you a bit of a tutorial of the data, you've got your IDs. 212 00:18:17,160 --> 00:18:23,540 You know what lab tested the cannabis. 213 00:18:23,540 --> 00:18:32,800 You've got the name that people attach to the inventory. 214 00:18:32,800 --> 00:18:36,680 As well as we now know what strain it is. 215 00:18:36,680 --> 00:18:45,060 And so you could do a incredible analysis of strains. 216 00:18:45,060 --> 00:18:54,280 So people have been really curious about what's in a strain, what's in Pangea, what's in train 217 00:18:54,280 --> 00:18:55,280 rack. 218 00:18:55,280 --> 00:19:02,660 So you can do an analysis here and try to find out what strains are people growing and 219 00:19:02,660 --> 00:19:05,640 which ones may look the most profitable. 220 00:19:05,640 --> 00:19:09,580 So that's a whole nother can of worms. 221 00:19:09,580 --> 00:19:13,360 But you can add a geographic layer. 222 00:19:13,360 --> 00:19:18,320 So maybe people are growing certain strains in certain parts of the state. 223 00:19:18,320 --> 00:19:24,600 So certain strains may be better in certain climates. 224 00:19:24,600 --> 00:19:27,880 You know what type of product it is. 225 00:19:27,880 --> 00:19:33,480 And so today we'll be specifically looking at concentrates. 226 00:19:33,480 --> 00:19:41,220 And then of course the gold, you have all of the cannabinoid data. 227 00:19:41,220 --> 00:19:44,820 So it would be awesome to have terpene data too. 228 00:19:44,820 --> 00:19:47,160 But we'll take what we're given. 229 00:19:47,160 --> 00:19:53,880 And in this case, we are given an awesome data set of cannabinoids. 230 00:19:53,880 --> 00:20:04,520 And you can also utilize data on residual solvents, microbes, mycotoxins. 231 00:20:04,520 --> 00:20:11,920 You know, just so that you can maybe see what other factors play into this, right? 232 00:20:11,920 --> 00:20:24,000 Because you also want to be growing strains that are, you know, handle stressors in your 233 00:20:24,000 --> 00:20:25,000 environment. 234 00:20:25,000 --> 00:20:30,320 So perhaps certain strains may be more likely to fail for micro or myco. 235 00:20:30,320 --> 00:20:31,320 I don't know. 236 00:20:31,320 --> 00:20:33,440 I'm just kind of conjecturing. 237 00:20:33,440 --> 00:20:40,080 But there could be many interesting aspects in the data. 238 00:20:40,080 --> 00:20:42,560 So there you have it. 239 00:20:42,560 --> 00:20:55,040 We've taken a 50 megabyte data set and flattened it down into 200 megabytes. 240 00:20:55,040 --> 00:21:01,600 So without further ado, I'll go ahead and jump into a bit of exploratory analysis. 241 00:21:01,600 --> 00:21:05,520 Are there any questions so far? 242 00:21:05,520 --> 00:21:08,960 Just checking real quick. 243 00:21:08,960 --> 00:21:10,520 I'm still with you all. 244 00:21:10,520 --> 00:21:11,520 All right. 245 00:21:11,520 --> 00:21:14,320 We'll must have you all entertained. 246 00:21:14,320 --> 00:21:19,160 So we'll get into some exploratory data analysis here, right? 247 00:21:19,160 --> 00:21:22,560 So we've got an awesome data set here. 248 00:21:22,560 --> 00:21:34,440 What's said and done, you know, we're looking at between 100,000 to, you know, 420,000 or 249 00:21:34,440 --> 00:21:37,200 so, 418,000. 250 00:21:37,200 --> 00:21:43,400 So observations, a lot of these may have missing values. 251 00:21:43,400 --> 00:21:46,960 And different people, so here's a good example. 252 00:21:46,960 --> 00:21:50,140 Different people have different naming conventions. 253 00:21:50,140 --> 00:21:58,800 So for example, this cultivator is really good about putting their strain name in, but 254 00:21:58,800 --> 00:22:03,960 they don't have any informative inventory name. 255 00:22:03,960 --> 00:22:17,200 Whereas other producers may have informative inventory names such as this producer. 256 00:22:17,200 --> 00:22:20,480 However, they've got less informative. 257 00:22:20,480 --> 00:22:24,760 Well, it's a give or take. 258 00:22:24,760 --> 00:22:35,520 So long story short, you need to look at both of these inventory names and strain names. 259 00:22:35,520 --> 00:22:37,920 But that's a whole other can of worms. 260 00:22:37,920 --> 00:22:40,640 We're going to get into cultivation next week. 261 00:22:40,640 --> 00:22:45,320 And this is definitely going to be something we look at. 262 00:22:45,320 --> 00:22:51,240 Because you know, the strain you grow, that's a critical choice. 263 00:22:51,240 --> 00:22:54,480 So we've got awesome lab result data there. 264 00:22:54,480 --> 00:22:59,600 And I'll make this available for the meetup group since we're all about open source, open 265 00:22:59,600 --> 00:23:02,600 data here. 266 00:23:02,600 --> 00:23:04,720 Cool. 267 00:23:04,720 --> 00:23:10,260 Well, let's get off to the races. 268 00:23:10,260 --> 00:23:13,920 So we don't need many tools for today. 269 00:23:13,920 --> 00:23:16,840 Probably just some of the basic Python functions. 270 00:23:16,840 --> 00:23:22,360 You know, well, just to recap, I was just saying, we're just going to start with some of these 271 00:23:22,360 --> 00:23:26,600 core packages, Matplotlib, pandas. 272 00:23:26,600 --> 00:23:28,840 You can go a long way with them. 273 00:23:28,840 --> 00:23:36,360 And I was just going to say, you can think of Seaborn as an extension of Matplotlib. 274 00:23:36,360 --> 00:23:40,680 So for me, it's easy to conceptualize it that way. 275 00:23:40,680 --> 00:23:44,360 That way, you just use Seaborn where Matplotlib falls short. 276 00:23:44,360 --> 00:23:47,160 But you know, that's sort of the nitty gritty. 277 00:23:47,160 --> 00:23:50,480 I use it for things like colors. 278 00:23:50,480 --> 00:24:00,160 And then quick tip is, if you just want nice looking plots, just throw everything in 538 279 00:24:00,160 --> 00:24:06,040 style and things look nice right out of the box. 280 00:24:06,040 --> 00:24:08,920 So I already have the data read in here. 281 00:24:08,920 --> 00:24:19,240 And here is a snippet that you'll be able to use once I share this data with you, where 282 00:24:19,240 --> 00:24:22,680 you can pick and choose which variables you want. 283 00:24:22,680 --> 00:24:30,280 So for example, if you don't need these variables, you can just comment them out and then read 284 00:24:30,280 --> 00:24:33,800 in the data that you do need. 285 00:24:33,800 --> 00:24:47,200 For example, the data I've read in, if you look at an observation here, we've got a big 286 00:24:47,200 --> 00:24:50,560 rich JSON data set here. 287 00:24:50,560 --> 00:25:01,080 So we have all the variables that I have already showed you in a nice JSON format. 288 00:25:01,080 --> 00:25:04,000 So awesome. 289 00:25:04,000 --> 00:25:08,120 Awesome, awesome, awesome. 290 00:25:08,120 --> 00:25:13,000 Well let's get to some of the interesting bits. 291 00:25:13,000 --> 00:25:23,240 So I realize now that we're also going to need to include, to say we're looking at processed 292 00:25:23,240 --> 00:25:25,040 products. 293 00:25:25,040 --> 00:25:35,040 You know, we'll also need to include products where they may have used solventless extractions. 294 00:25:35,040 --> 00:25:50,320 So let's see if we can't identify some of these lab results where they use solventless. 295 00:25:50,320 --> 00:25:58,520 So the way we can do this is first we can remember, let's look at some of these fields 296 00:25:58,520 --> 00:26:01,000 that we have here. 297 00:26:01,000 --> 00:26:06,160 So we've got inventory name, not quite what we need. 298 00:26:06,160 --> 00:26:08,160 We need the inventory type. 299 00:26:08,160 --> 00:26:12,440 I think we may even want sample. 300 00:26:12,440 --> 00:26:17,160 Here we are, intermediate type. 301 00:26:17,160 --> 00:26:27,360 Let's just list out here all of the different sample types that we have. 302 00:26:27,360 --> 00:26:32,640 So here they are. 303 00:26:32,640 --> 00:26:35,120 Today not going to be looking at flour. 304 00:26:35,120 --> 00:26:39,460 So next week we'll look exclusively at flour. 305 00:26:39,460 --> 00:26:44,600 For today, just kind of wanted to do a deep dive into concentrates. 306 00:26:44,600 --> 00:26:47,720 That way we're just comparing apples to apples. 307 00:26:47,720 --> 00:26:53,840 So let's see if we can't just say, okay, let's get everything where they either did a solvent 308 00:26:53,840 --> 00:27:10,560 test, or we've got these intermediate types that doesn't require 309 00:27:10,560 --> 00:27:15,960 solvent screening. 310 00:27:15,960 --> 00:27:23,880 So I believe if you're using a food grade solvent, then you don't need to get residual 311 00:27:23,880 --> 00:27:25,440 solvent screening. 312 00:27:25,440 --> 00:27:29,920 It should be included in the or statement nonetheless. 313 00:27:29,920 --> 00:27:36,920 And then of course you don't need residual solvent screening if you're using a non solventless 314 00:27:36,920 --> 00:27:39,840 extraction method. 315 00:27:39,840 --> 00:27:43,120 So I may have missed some. 316 00:27:43,120 --> 00:27:48,600 So this may exclude products such as edibles. 317 00:27:48,600 --> 00:27:58,320 We may want to repeat this analysis specifically for edibles to look at the average concentration 318 00:27:58,320 --> 00:27:59,720 in edibles. 319 00:27:59,720 --> 00:28:07,160 However, just going to say, I don't believe there's as much variability in edibles due 320 00:28:07,160 --> 00:28:09,840 to the regulations on edibles. 321 00:28:09,840 --> 00:28:17,920 I do believe there's a limit on the milligrams of THC and CBD that can be in an individual 322 00:28:17,920 --> 00:28:18,920 edible. 323 00:28:18,920 --> 00:28:28,040 So you see, for example, if the limit is just picking a number here, 10 milligrams per gram, 324 00:28:28,040 --> 00:28:33,680 then you're just going to see a lot of people trying to produce exactly 10. 325 00:28:33,680 --> 00:28:36,560 So it's worth a deep dive. 326 00:28:36,560 --> 00:28:43,280 And they just going to try to work here on processed products. 327 00:28:43,280 --> 00:28:48,840 So we basically are just going to start doing a bunch of conditional averages. 328 00:28:48,840 --> 00:28:53,220 I say you can go so far with that. 329 00:28:53,220 --> 00:28:54,620 It's extraordinary. 330 00:28:54,620 --> 00:29:01,840 It's almost like you just kind of keep layering on conditions and the data just becomes more 331 00:29:01,840 --> 00:29:04,280 and more interesting. 332 00:29:04,280 --> 00:29:12,480 So let's just once again, let's just look at our intermediate types here just to ensure 333 00:29:12,480 --> 00:29:16,600 that we didn't get anything odd in the mix. 334 00:29:16,600 --> 00:29:17,840 And we did. 335 00:29:17,840 --> 00:29:25,680 We've got some flowers and some products we couldn't identify here. 336 00:29:25,680 --> 00:29:30,200 So let's just see if we can't just exclude those real quick. 337 00:29:30,200 --> 00:29:49,280 So let's just say we want all the products where we don't have, let's just say everything 338 00:29:49,280 --> 00:29:56,840 where the intermediate type is there. 339 00:29:56,840 --> 00:30:00,600 We don't want to look at flowers for today. 340 00:30:00,600 --> 00:30:03,520 That's probably just a miscoding there. 341 00:30:03,520 --> 00:30:09,840 So let's see if we can refine our products a little better. 342 00:30:09,840 --> 00:30:14,400 OK, so we can't negate that. 343 00:30:14,400 --> 00:30:16,480 So we'll negate there. 344 00:30:16,480 --> 00:30:19,480 That's better. 345 00:30:19,480 --> 00:30:22,920 Awesome. 346 00:30:22,920 --> 00:30:32,560 So let's just double check that we just have. 347 00:30:32,560 --> 00:30:41,920 Well, somehow it looks like you may just have to proceed as is. 348 00:30:41,920 --> 00:30:45,760 That's for expediency's sake here. 349 00:30:45,760 --> 00:30:48,480 OK, so not getting those. 350 00:30:48,480 --> 00:31:08,800 Aha, just need to fix this to an and condition. 351 00:31:08,800 --> 00:31:17,040 We may not be able to get rid of the nans for now, but that's OK. 352 00:31:17,040 --> 00:31:23,080 We're able to at least exclude the flowers. 353 00:31:23,080 --> 00:31:24,080 Question? 354 00:31:24,080 --> 00:31:29,560 OK, OK, Dean, apologize though. 355 00:31:29,560 --> 00:31:36,000 It is a bit, but good to have you nonetheless. 356 00:31:36,000 --> 00:31:37,560 And sorry for all that. 357 00:31:37,560 --> 00:31:39,520 It may be a little scattered today. 358 00:31:39,520 --> 00:31:46,000 I spent most of my time just writing this big script to compile all the data. 359 00:31:46,000 --> 00:31:49,980 So next week, I'll try to have things a bit more organized. 360 00:31:49,980 --> 00:31:53,760 So that's why it's sort of just been from the top today. 361 00:31:53,760 --> 00:31:59,640 So my apologies, but we'll hone things in through the future. 362 00:31:59,640 --> 00:32:04,200 But without further ado, let's get to some interesting numbers and some figures here. 363 00:32:04,200 --> 00:32:09,500 We should have gotten to the figures before Dean had to go. 364 00:32:09,500 --> 00:32:16,080 So we'll go ahead and get the total cannabinoids here. 365 00:32:16,080 --> 00:32:18,520 Interesting. 366 00:32:18,520 --> 00:32:42,640 It looks like we may have missed some of these cannabinoids. 367 00:32:42,640 --> 00:32:45,880 Interesting. 368 00:32:45,880 --> 00:32:49,860 So it looks like we may have missed some of these cannabinoids. 369 00:32:49,860 --> 00:32:57,960 So I think I may have missed CBN. 370 00:32:57,960 --> 00:33:00,480 You've missed some of these. 371 00:33:00,480 --> 00:33:01,480 Interesting. 372 00:33:01,480 --> 00:33:10,920 Oh, well, we'll proceed with the data that we do have. 373 00:33:10,920 --> 00:33:17,000 OK, also don't have CBGA and THCV. 374 00:33:17,000 --> 00:33:18,000 So my apologies. 375 00:33:18,000 --> 00:33:23,800 I'll have to add those cannabinoids back to the data set. 376 00:33:23,800 --> 00:33:26,800 Awesome. 377 00:33:26,800 --> 00:33:41,560 So we now are down to a small data set here, only have about 800 observations. 378 00:33:41,560 --> 00:33:49,600 So want to relook at why that may be. 379 00:33:49,600 --> 00:33:55,160 So like I said, this is sort of the first go at analyzing this data. 380 00:33:55,160 --> 00:33:59,480 The different types of concentrates you see in the market. 381 00:33:59,480 --> 00:34:03,360 So this is where things actually get interesting. 382 00:34:03,360 --> 00:34:12,680 So we can see, wow, a bulk of the market is done with ethanol concentrates, followed by 383 00:34:12,680 --> 00:34:23,560 hydrocarbon, followed by CO2, with food grade and non-solvent lists making up just above 384 00:34:23,560 --> 00:34:27,200 10% of the whole market. 385 00:34:27,200 --> 00:34:31,360 So any thoughts or comments about this breakdown? 386 00:34:31,360 --> 00:34:32,360 Question? 387 00:34:32,360 --> 00:34:38,360 Yeah, for the solvent versus non-solvent base, would the non-solvent be a lot more expensive? 388 00:34:38,360 --> 00:34:40,160 I'm not too familiar with the processes. 389 00:34:40,160 --> 00:34:41,720 Is that why it would be lower? 390 00:34:41,720 --> 00:34:43,880 At least one of the reasons. 391 00:34:43,880 --> 00:34:46,840 Well, you raise a real interesting point. 392 00:34:46,840 --> 00:34:51,440 And so this is something that depends. 393 00:34:51,440 --> 00:35:00,200 So at the retail level, do consumers have a demand for non-solvent list extracts? 394 00:35:00,200 --> 00:35:02,880 So that's an awesome question. 395 00:35:02,880 --> 00:35:11,600 So we should combine this with retail data and see, does two products that are made with 396 00:35:11,600 --> 00:35:19,120 solvent list methods, do they retail at a higher price than other concentrates? 397 00:35:19,120 --> 00:35:24,320 So is there a premium place on non-solvent list? 398 00:35:24,320 --> 00:35:27,440 And it also may depend on the concentration. 399 00:35:27,440 --> 00:35:35,120 So I've been wanting to do an analysis on, does concentration of cannabinoids affect 400 00:35:35,120 --> 00:35:38,160 the retail price? 401 00:35:38,160 --> 00:35:41,200 I don't know if anybody's looked at that yet. 402 00:35:41,200 --> 00:35:46,080 And I think it would be so easy to do with this data that we have here in Washington. 403 00:35:46,080 --> 00:35:54,160 All we would do is run a regression of price, which we have. 404 00:35:54,160 --> 00:35:55,160 We're at sales. 405 00:35:55,160 --> 00:36:03,400 We have sales on the concentration of that product and see if there's a positive or negative 406 00:36:03,400 --> 00:36:04,400 correlation. 407 00:36:04,400 --> 00:36:11,680 All we're going to do is add a time component to our analysis. 408 00:36:11,680 --> 00:36:20,200 And then we're essentially going to track the breakdown of analyses used over time. 409 00:36:20,200 --> 00:36:29,620 So is solvent list becoming a greater or a smaller proportion of the market? 410 00:36:29,620 --> 00:36:32,320 We don't know, but we can easily answer that. 411 00:36:32,320 --> 00:36:35,960 Did that answer your question? 412 00:36:35,960 --> 00:36:39,120 Yeah, for the most part. 413 00:36:39,120 --> 00:36:43,200 I'm wondering about price and also if it's a lot more difficult to do, which I'm assuming 414 00:36:43,200 --> 00:36:44,760 it would be. 415 00:36:44,760 --> 00:36:45,760 Exactly. 416 00:36:45,760 --> 00:36:54,200 If you're doing a high pressure non-solvent list extraction, that may be challenging. 417 00:36:54,200 --> 00:36:55,760 There are people out there who are doing it. 418 00:36:55,760 --> 00:36:59,000 And then yes, the price point is critical. 419 00:36:59,000 --> 00:37:01,240 So let's look at it. 420 00:37:01,240 --> 00:37:06,360 Let's essentially combine this data with the price data. 421 00:37:06,360 --> 00:37:10,440 And like I said, it would just be interesting to see if there's any correlation between 422 00:37:10,440 --> 00:37:14,280 price and any of the other factors. 423 00:37:14,280 --> 00:37:20,960 Because it doesn't necessarily matter. 424 00:37:20,960 --> 00:37:25,000 It all depends on consumers' preferences. 425 00:37:25,000 --> 00:37:31,480 Too bad one of our Classic members isn't here today because she could talk about her preferences 426 00:37:31,480 --> 00:37:38,360 because consumers are hard to predict. 427 00:37:38,360 --> 00:37:41,320 So it's really hard to predict what they'd like. 428 00:37:41,320 --> 00:37:44,360 But we can look at the data. 429 00:37:44,360 --> 00:37:47,040 Awesome. 430 00:37:47,040 --> 00:37:52,520 Just to round it out here, we can start looking at some of the concentrations. 431 00:37:52,520 --> 00:37:57,960 So this is just sort of pseudo code. 432 00:37:57,960 --> 00:38:04,960 So still need to make sure that everything works correctly. 433 00:38:04,960 --> 00:38:15,720 But here we have the average concentration and the standard deviation found in various 434 00:38:15,720 --> 00:38:16,960 concentrates. 435 00:38:16,960 --> 00:38:24,280 So let's just look at the average concentration first. 436 00:38:24,280 --> 00:38:31,780 So here you have, topping the charts, this actually kind of surprises me. 437 00:38:31,780 --> 00:38:41,920 Topping the charts, you have ethanol coming in at an average of 81% total cannabinoids 438 00:38:41,920 --> 00:38:51,400 with hydrocarbon concentrates just behind at an average of 80%. 439 00:38:51,400 --> 00:39:03,160 And then quite a lower average, you've got CO2 concentrates and food grade solvents around 440 00:39:03,160 --> 00:39:08,200 60-61% for CO2. 441 00:39:08,200 --> 00:39:10,320 I find that interesting. 442 00:39:10,320 --> 00:39:19,360 And then here you see the non-solvent list actually has the lowest average cannabinoids. 443 00:39:19,360 --> 00:39:25,120 And I would conjecture that that's because I wouldn't be surprised if people who are 444 00:39:25,120 --> 00:39:34,840 doing non-solvent list extracts may be doing things like bubble hash, which I don't think 445 00:39:34,840 --> 00:39:44,120 necessarily will have the extraction. 446 00:39:44,120 --> 00:39:51,200 I'm not sure what the word is for it, efficiency perhaps, that some of the other extraction 447 00:39:51,200 --> 00:39:52,200 may- 448 00:39:52,200 --> 00:39:53,200 No, that makes sense. 449 00:39:53,200 --> 00:39:55,960 Do you think more product would have to be used because it's less efficient and that 450 00:39:55,960 --> 00:40:00,400 might be part of the high enterprise, I guess? 451 00:40:00,400 --> 00:40:07,000 Well, that's what I have in my prior is that it does. 452 00:40:07,000 --> 00:40:16,680 So my prior is that consumers have a preference for higher cannabinoid concentrations. 453 00:40:16,680 --> 00:40:19,280 So you'd get a price premium, right? 454 00:40:19,280 --> 00:40:27,560 They're basically paying per milligram of THC. 455 00:40:27,560 --> 00:40:37,960 So you'd have to sell something that's 53% at a lower price than something that's 80% 456 00:40:37,960 --> 00:40:47,240 THC for the consumer to purchase the same milligrams of THC. 457 00:40:47,240 --> 00:40:54,920 So if you abstracted that, oh, the consumer's in the market for milligrams of THC, then 458 00:40:54,920 --> 00:41:01,600 you would expect that something with 53% would have a lower price point. 459 00:41:01,600 --> 00:41:09,080 If it's got a lower price point, there may not be as many people in the market for it. 460 00:41:09,080 --> 00:41:22,960 But I think this is critical to look at the strain name on this slide because I think 461 00:41:22,960 --> 00:41:28,880 the strain and inventory name could be really informative because you may want to try to 462 00:41:28,880 --> 00:41:35,320 separate out if there's different types of these non-solventless extracts. 463 00:41:35,320 --> 00:41:42,160 So for example, some may specifically say there's bubble hatch and others may not. 464 00:41:42,160 --> 00:41:48,520 And it would be interesting to see if there's any groupings that you can make within these 465 00:41:48,520 --> 00:41:53,600 product categories. 466 00:41:53,600 --> 00:42:00,880 And then basically, just to finish out what we began with, essentially I was saying, OK, 467 00:42:00,880 --> 00:42:05,880 you've got your different extraction mechanisms and they're going to have their different 468 00:42:05,880 --> 00:42:08,400 efficacy rates. 469 00:42:08,400 --> 00:42:16,880 If you have two extraction mechanisms that are about the same, so that would be hydrocarbon 470 00:42:16,880 --> 00:42:24,080 concentrates and ethanol concentrates, and as well CO2 concentrates and food grade solvent 471 00:42:24,080 --> 00:42:33,000 concentrates, then from these reasons that we stated in the presentation, from financial 472 00:42:33,000 --> 00:42:39,200 reasons, economic reasons, engineering, business marketing, public health, you would gravitate 473 00:42:39,200 --> 00:42:48,200 towards the one that has a lower variance, lower standard deviation. 474 00:42:48,200 --> 00:42:52,040 So here, let's see if we can't pick the ones that would win. 475 00:42:52,040 --> 00:43:04,160 So if we're comparing hydrocarbons to ethanol extraction, ethanol just takes the cake. 476 00:43:04,160 --> 00:43:10,600 So hydrocarbon concentrates, and like I said, we want to look at this like a processor by 477 00:43:10,600 --> 00:43:19,760 processor basis and even see if we can't disentangle these into more subgroupings. 478 00:43:19,760 --> 00:43:27,920 But right out of the bat, our first impression is, oh, hydrocarbon concentrates have standard 479 00:43:27,920 --> 00:43:41,640 deviation of almost 10%, whereas ethanol concentrates by and far have the lowest variability, the 480 00:43:41,640 --> 00:43:47,980 lowest standard deviation, but their standard deviation being less than 4%. 481 00:43:47,980 --> 00:44:01,480 So that means you would expect around 95% of your products would be within 8% of 81%. 482 00:44:01,480 --> 00:44:14,880 So you would expect your products, they could be varying from around 73 to around 90% concentrate. 483 00:44:14,880 --> 00:44:20,360 And this is why you could also look at it by a processor by processor basis, which is 484 00:44:20,360 --> 00:44:27,680 exactly what we'll be doing on Saturday, because we can do this exact same analysis processor 485 00:44:27,680 --> 00:44:31,400 by processor and over time. 486 00:44:31,400 --> 00:44:39,320 And what that would show would be, okay, are the processors who are testing the most samples, 487 00:44:39,320 --> 00:44:49,560 do they have a high concentration and do they have a low standard deviation in their concentration? 488 00:44:49,560 --> 00:45:00,280 And I would naively predict that there's a positive correlation between average concentration 489 00:45:00,280 --> 00:45:07,880 and the total number of samples you test, as well as a negative correlation between 490 00:45:07,880 --> 00:45:12,560 your standard deviation in the samples you test. 491 00:45:12,560 --> 00:45:20,000 And so I would just basically say the people who are able to operate with the lower variability, 492 00:45:20,000 --> 00:45:24,480 I conjecture they're going to be more profitable. 493 00:45:24,480 --> 00:45:31,320 We'll let the data show on Saturday. 494 00:45:31,320 --> 00:45:37,880 And also, you know, we can look at this over time to see, okay, are people honing in their 495 00:45:37,880 --> 00:45:41,880 processes over time? 496 00:45:41,880 --> 00:45:47,120 Are people's standard deviations, is this decreasing over time? 497 00:45:47,120 --> 00:45:49,120 Is it increasing? 498 00:45:49,120 --> 00:45:51,200 How about concentration? 499 00:45:51,200 --> 00:45:57,360 Is concentration increasing, decreasing, staying constant? 500 00:45:57,360 --> 00:46:07,000 Is for example, if concentration starts to become pretty stable, everybody sort of, they've 501 00:46:07,000 --> 00:46:08,480 got their extraction techniques. 502 00:46:08,480 --> 00:46:10,440 That may not be the case. 503 00:46:10,440 --> 00:46:13,600 People may be getting better and better extraction techniques over time. 504 00:46:13,600 --> 00:46:16,560 Have a standard deviation of almost 18%. 505 00:46:16,560 --> 00:46:23,040 I was just going to say rounding out for today, if you were going to compare food grade solvents 506 00:46:23,040 --> 00:46:29,760 with CO2 concentrates, you would look at the standard deviation since their average concentration 507 00:46:29,760 --> 00:46:31,840 is about the same. 508 00:46:31,840 --> 00:46:38,640 And you would see people with who are producing with food grade solvent concentrates on average 509 00:46:38,640 --> 00:46:46,160 have a standard deviation of almost 18%, which is when you think about it, quite extraordinary. 510 00:46:46,160 --> 00:46:54,040 I would suggest, you know, it wouldn't be implausible to have your food grade concentrate, 511 00:46:54,040 --> 00:47:04,520 you know, almost 100% as well as almost 10% or 15%. 512 00:47:04,520 --> 00:47:07,120 So would that actually shake out? 513 00:47:07,120 --> 00:47:08,120 May or may not. 514 00:47:08,120 --> 00:47:15,440 But long story short, that's quite a, that's quite a high, quite a high variation. 515 00:47:15,440 --> 00:47:22,680 Same with the non solvent list, the people who are doing non solvent lists, big variation. 516 00:47:22,680 --> 00:47:31,560 So if, you know, if I were to suggest an extraction method, you know, well, I may not suggest 517 00:47:31,560 --> 00:47:32,560 this one. 518 00:47:32,560 --> 00:47:38,280 So, you know, so if you were going to do food grade solvent versus CO2 concentrate, you 519 00:47:38,280 --> 00:47:44,600 know, it does look like, you know, CO2 concentrates have about the same, if not higher average 520 00:47:44,600 --> 00:47:48,520 concentration. 521 00:47:48,520 --> 00:47:53,160 And they also have a lower standard deviation. 522 00:47:53,160 --> 00:47:56,800 Keep in mind, it matters what you measure. 523 00:47:56,800 --> 00:48:04,120 So remember here, we still have a whole lot of other factors to look at. 524 00:48:04,120 --> 00:48:08,520 So in particular, the solvents. 525 00:48:08,520 --> 00:48:16,680 So I would want to know, okay, our CO2, it's got it around the same concentration, it's 526 00:48:16,680 --> 00:48:19,280 got lower standard deviation. 527 00:48:19,280 --> 00:48:21,520 What's its failure rate? 528 00:48:21,520 --> 00:48:28,200 Because you know, if we're failing for solvents, left and right, that's not going to necessarily 529 00:48:28,200 --> 00:48:30,020 be profitable. 530 00:48:30,020 --> 00:48:38,200 So this is why, you know, running a profitable business, it just matters on so many metrics, 531 00:48:38,200 --> 00:48:42,480 on so many fronts, right, you've got to worry about your concentration, you've got to worry 532 00:48:42,480 --> 00:48:48,480 about your standard deviation, you've got to worry about your failure rate. 533 00:48:48,480 --> 00:48:52,760 You know, and for example, what are consumers preferences? 534 00:48:52,760 --> 00:48:59,560 So that's why it'd be interesting to see what the shakedown is in the market over time. 535 00:48:59,560 --> 00:49:07,640 So are consumers gravitating towards one type of product or another? 536 00:49:07,640 --> 00:49:18,200 You know, it's just just so many factors here that play out in running a profitable producer, 537 00:49:18,200 --> 00:49:20,840 processor. 538 00:49:20,840 --> 00:49:26,680 And I think the lesson of the day is, you know, variability matters. 539 00:49:26,680 --> 00:49:34,320 So you know, just if you're looking at variability, you can start to make educated decisions of, 540 00:49:34,320 --> 00:49:39,920 okay, do I choose hydrocarbon concentrate, do I choose ethanol? 541 00:49:39,920 --> 00:49:42,880 And like I said, this isn't the end of the story. 542 00:49:42,880 --> 00:49:46,760 You'll want to see if you can't make subgroups. 543 00:49:46,760 --> 00:49:57,560 So see, okay, can we use the strain name and inventory name to even further classify these 544 00:49:57,560 --> 00:49:59,240 samples? 545 00:49:59,240 --> 00:50:06,080 And then also, of course, you'll want to, you know, take a look at some of the variables 546 00:50:06,080 --> 00:50:11,920 that you're screening for just to make sure that, you know, everything's passing at the 547 00:50:11,920 --> 00:50:15,880 rate that you want it to. 548 00:50:15,880 --> 00:50:22,200 And for that, I'm going to go ahead and conclude the presentation just so I'm respectful of 549 00:50:22,200 --> 00:50:23,680 everybody's time. 550 00:50:23,680 --> 00:50:26,040 However, were there any questions? 551 00:50:26,040 --> 00:50:31,760 I know it was a bit of a rocky start, but hopefully we got into gear there towards the 552 00:50:31,760 --> 00:50:32,760 end. 553 00:50:32,760 --> 00:50:34,760 So any questions? 554 00:50:34,760 --> 00:50:36,040 Comment? 555 00:50:36,040 --> 00:50:42,000 I guess just for the forecasting you're talking about on Saturday, you guys usually just run 556 00:50:42,000 --> 00:50:45,880 like simple regressions, pretty much. 557 00:50:45,880 --> 00:50:53,600 We heavily lean on Fox Jenkins slash Arima forecasting, if you may have heard of it. 558 00:50:53,600 --> 00:50:59,520 So it's essentially a time series technique. 559 00:50:59,520 --> 00:51:03,120 You just use an observation that you can track over time. 560 00:51:03,120 --> 00:51:08,040 In this case, we'll just be tracking concentrations over time month by month. 561 00:51:08,040 --> 00:51:10,040 You could potentially do day by day. 562 00:51:10,040 --> 00:51:15,900 I like month by month or even week by week, week by week is quite powerful. 563 00:51:15,900 --> 00:51:23,720 And then you just use a series of observations and you forecast them into the future. 564 00:51:23,720 --> 00:51:29,320 So you basically use things like, oh, does this series have a trend? 565 00:51:29,320 --> 00:51:33,280 Is there a moving average component? 566 00:51:33,280 --> 00:51:36,840 And then you basically, it is a regression model. 567 00:51:36,840 --> 00:51:40,600 It is a regression model at the end of the day. 568 00:51:40,600 --> 00:51:47,720 And yes, you can get quite accurate, surprisingly accurate forecasts for the future. 569 00:51:47,720 --> 00:51:56,360 So I always like to say it's a heck of a lot better having a guesstimate of the future 570 00:51:56,360 --> 00:52:00,040 than no estimate of the future. 571 00:52:00,040 --> 00:52:08,400 So definitely join us if you want, because not only will we be forecasting the average 572 00:52:08,400 --> 00:52:17,200 concentration for aggregate and by individual processor, we'll also get to forecast their 573 00:52:17,200 --> 00:52:25,640 variability, which is a more advanced statistical technique, but we'll do it simply and easily 574 00:52:25,640 --> 00:52:28,360 because that's the way things should be done. 575 00:52:28,360 --> 00:52:32,060 And we can get a nice forecast for people's variability. 576 00:52:32,060 --> 00:52:39,440 And that way we can rank order all of the processors and see, OK, who's got the highest 577 00:52:39,440 --> 00:52:41,680 average concentration week by week? 578 00:52:41,680 --> 00:52:46,280 Who's got the lowest standard deviation week by week? 579 00:52:46,280 --> 00:52:51,140 And then see how the industry is moving. 580 00:52:51,140 --> 00:52:56,180 So is there any trend in average concentrations? 581 00:52:56,180 --> 00:52:58,920 Are people getting more efficient over time? 582 00:52:58,920 --> 00:53:00,840 We'll be able to answer all those questions. 583 00:53:00,840 --> 00:53:02,440 It will be exciting. 584 00:53:02,440 --> 00:53:04,480 OK, thank you. 585 00:53:04,480 --> 00:53:05,480 Definitely. 586 00:53:05,480 --> 00:53:09,360 And if you can't make it on Saturday, just feel free to register. 587 00:53:09,360 --> 00:53:13,560 It's $1 and I'll send you all the source code and material. 588 00:53:13,560 --> 00:53:18,240 So I hope your value will be tenfold, if not more. 589 00:53:18,240 --> 00:53:22,360 So check it out, Saturday morning statistics. 590 00:53:22,360 --> 00:53:28,320 If not, then next Wednesday we'll pick up with producer cultivators. 591 00:53:28,320 --> 00:53:33,280 So we'll look more at the cultivation side next time, maybe kind of recap on what we 592 00:53:33,280 --> 00:53:34,760 did today. 593 00:53:34,760 --> 00:53:37,160 So it's extraordinary work. 594 00:53:37,160 --> 00:53:44,600 So if you have any breakthrough discoveries, questions along the way, ideas, feel free 595 00:53:44,600 --> 00:53:50,320 to share them because it would be awesome to have a collaborative effort to really dig 596 00:53:50,320 --> 00:53:57,840 deep into this data because there is so many questions to answer in not enough time. 597 00:53:57,840 --> 00:53:58,840 We'll do it. 598 00:53:58,840 --> 00:54:03,880 We'll do a lot though. 599 00:54:03,880 --> 00:54:05,100 Thank you all for coming. 600 00:54:05,100 --> 00:54:07,760 So I can't thank you enough for coming. 601 00:54:07,760 --> 00:54:10,960 So it's been fun. 602 00:54:10,960 --> 00:54:16,960 And hopefully we were able to get through some of the internet lags there and maybe 603 00:54:16,960 --> 00:54:18,560 you found it worthwhile. 604 00:54:18,560 --> 00:54:20,680 So definitely feel free to share some of your feedback. 605 00:54:20,680 --> 00:54:24,320 I'm always here if you want to send a message. 606 00:54:24,320 --> 00:54:25,880 Awesome. 607 00:54:25,880 --> 00:54:26,880 Thank you, Deacon. 608 00:54:26,880 --> 00:54:27,880 Awesome. 609 00:54:27,880 --> 00:54:28,880 Thank you all. 610 00:54:28,880 --> 00:54:56,960 Let's get in stone and have an awesome week!